prithivMLmods/Galactic-Qwen-14B-Exp2
prithivMLmods/Galactic-Qwen-14B-Exp2 is a 14.8 billion parameter model based on the Qwen 2.5 architecture, fine-tuned to enhance reasoning capabilities. It excels in contextual understanding, logical deduction, and multi-step problem-solving, optimized through a long chain-of-thought reasoning model. The model supports a 131072-token context length and is proficient in over 29 languages, making it suitable for general-purpose reasoning and multilingual applications.
Loading preview...
Galactic-Qwen-14B-Exp2 Overview
Galactic-Qwen-14B-Exp2 is a 14.8 billion parameter model built on the Qwen 2.5 architecture, specifically designed to boost the reasoning abilities of 14B-parameter models. It has been fine-tuned using a long chain-of-thought reasoning model and specialized datasets to improve comprehension, structured responses, and conversational intelligence.
Key Improvements & Capabilities
- Enhanced General Knowledge: Provides broad knowledge across various domains for accurate and coherent responses.
- Improved Instruction Following: Advanced understanding and execution of complex instructions, generating structured and coherent outputs.
- Versatile Adaptability: Resilient to diverse prompts, handling a wide range of topics and conversation styles.
- Long-Context Support: Supports up to 128K tokens for input context and can generate up to 8K tokens in a single output.
- Multilingual Proficiency: Supports over 29 languages, including English, Chinese, French, Spanish, and more.
Intended Use Cases
- General-Purpose Reasoning: Ideal for logical reasoning, diverse question answering, and general knowledge problems.
- Educational and Informational Assistance: Suitable for providing explanations, summaries, and research-based responses.
- Conversational AI and Chatbots: Excellent for building intelligent agents requiring contextual understanding.
- Multilingual Applications: Supports global communication, translations, and multilingual content generation.
- Structured Data Processing: Capable of analyzing and generating structured outputs like tables and JSON.
- Long-Form Content Generation: Can produce extended responses such as articles, reports, and guides while maintaining coherence.
Performance Metrics
Evaluations on the Open LLM Leaderboard show an Average score of 43.56%, with specific results including:
- BBH (3-Shot): 59.92%
- MMLU-PRO (5-shot): 52.12%
- IFEval (0-Shot): 66.20%
For detailed results, refer to the Open LLM Leaderboard Evaluation Results.